opennn::GrowingNeurons Class Reference

Selects the optimal hidden neuron count by incrementally growing the number of neurons. More...

#include <growing_neurons.h>

Inheritance diagram for opennn::GrowingNeurons:

Public Member Functions
	GrowingNeurons (TrainingStrategy *=nullptr)
	Constructs the algorithm bound to an optional training strategy.
void	set_default ()
	Restores default search bounds and stopping criteria.
void	set_neurons_increment (const Index)
	Sets the step size used when growing the number of neurons between trials.
NeuronsSelectionResults	perform_neurons_selection () override
	Runs the neuron growing procedure until the stopping criterion is met.
void	from_JSON (const JsonDocument &) override
	Loads algorithm configuration from a JSON document.
void	to_JSON (JsonWriter &) const override
	Writes algorithm configuration to a JSON writer.
Public Member Functions inherited from opennn::NeuronSelection
	NeuronSelection (TrainingStrategy *=nullptr)
	Constructs the algorithm bound to an optional training strategy.
virtual	~NeuronSelection ()=default
const TrainingStrategy *	get_training_strategy () const
bool	has_training_strategy () const
bool	get_display () const
void	set (TrainingStrategy *)
	Binds the algorithm to the given training strategy.
void	set_training_strategy (TrainingStrategy *new_training_strategy)
void	set_default ()
	Restores default search bounds and stopping criteria.
void	set_maximum_neurons (const Index new_maximum_neurons)
void	set_minimum_neurons (const Index new_minimum_neurons)
void	set_trials_number (const Index new_trials_number)
void	set_display (bool new_display)
void	set_validation_error_goal (const float new_validation_error_goal)
void	set_maximum_epochs (const Index new_maximum_epochs)
void	set_maximum_validation_failures (const Index new_maximum_validation_failures)
void	set_maximum_time (const float new_maximum_time)
void	check () const
	Verifies that the training strategy and its dependencies are valid for neurons selection.
string	get_name () const
void	save (const filesystem::path &) const
	Saves the algorithm configuration to disk.
void	load (const filesystem::path &)
	Loads the algorithm configuration from disk.
virtual void	print () const
	Prints a human-readable description of the algorithm to stdout.

Additional Inherited Members
Public Types inherited from opennn::NeuronSelection
enum class	StoppingCondition {   MaximumTime , SelectionErrorGoal , MaximumEpochs , MaximumSelectionFailures ,   MaximumNeurons }
	Reasons the neurons selection loop may terminate. More...
Protected Attributes inherited from opennn::NeuronSelection
TrainingStrategy *	training_strategy = nullptr
VectorR	validation_error_history
VectorR	training_error_history
Index	minimum_neurons = 0
Index	maximum_neurons = 0
Index	trials_number = 1
float	validation_error_goal = 0
Index	maximum_epochs = 10
Index	maximum_validation_failures = 100
float	maximum_time = 0
bool	display = true
string	name

Detailed Description

Selects the optimal hidden neuron count by incrementally growing the number of neurons.

Constructor & Destructor Documentation

GrowingNeurons()

opennn::GrowingNeurons::GrowingNeurons

(

TrainingStrategy *

= nullptr

)

Constructs the algorithm bound to an optional training strategy.

Member Function Documentation

from_JSON()

void opennn::GrowingNeurons::from_JSON ( const JsonDocument & )

overridevirtual

Loads algorithm configuration from a JSON document.

Implements opennn::NeuronSelection.

perform_neurons_selection()

NeuronsSelectionResults opennn::GrowingNeurons::perform_neurons_selection ( )

overridevirtual

Runs the neuron growing procedure until the stopping criterion is met.

Returns

Selection results including the optimal neuron count and error history.

Implements opennn::NeuronSelection.

set_default()

void opennn::GrowingNeurons::set_default

(

)

Restores default search bounds and stopping criteria.

set_neurons_increment()

void opennn::GrowingNeurons::set_neurons_increment

(

const Index

)

Sets the step size used when growing the number of neurons between trials.

to_JSON()

void opennn::GrowingNeurons::to_JSON ( JsonWriter & ) const

overridevirtual

Writes algorithm configuration to a JSON writer.

Implements opennn::NeuronSelection.